The Research On The Model And Algorithm Of Bandwidth Measurement In IP Network

As Internet growing rapidly, more and more network applications need to know the network bandwidth information to monitor network utilization and performance. Knowledge of bandwidth is critical for numerous important network management tasks, including usage-based accounting, traffic engineering, QoS monitoring, and attack/intrusion detection. Yet the importance of bandwidth measurement capabilities is compounded by the fact that IP networks do not maintain per-flow state. In a sense, the scalability of the stateless IP networks has been bought at the expense of observables, especially in large scale.Recently, many monitoring tools have been developed to monitor network bandwidth. In general, conventional monitoring schemes to measure network bandwidth are classified into two types:Active measurement monitors network traffic by sending probe packets. This method uses the probe-packet stream to determine the network traffic indirectly. This means that we implicitly assume that performance of networks is the same as the values measured from active probe packets. Up to the present these active monitoring schemes will result in significant volumes of additional network traffic. The overhead that schemes impose on the underlying router can be significant and can adversely impact the router's throughput when used extensively.Passive measurement is usually monitor network traffic at some special position, which not imposes extra traffic or modifies packet. This means that this method will not affect network traffic. Yet we need extra schemes to gather and harmonize data from the distributed monitoring elements, these schemes are extensively focused on efficient monitoring with reducing the generation overhead communication as more as possible.In this paper, we make research on the deployment of monitors and propose several models with considering different active and passive measurement scheme.1. In the passive measurement research, there are fewer members in the monitor set the overhead for monitoring the network is not high. The problem of efficient monitoring is...